Vinod B
Published in

Vinod B

The Electoral College is a National Security Problem

The election in 2016 saw incredibly close margins in some states with margins of less than one percent. Given the relatively small electoral vote difference in the election (Trump and Clinton won states with 306 and 232 electoral votes respectively for a difference of only 74), these close states were quite pivotal. Changing how just a few people voted could have changed the results.

Indeed elections today are getting closer and closer. The popular vote margin over time has been steadily declining as the plot below shows.

In addition, states themselves are becoming more polarized and voting consistently for one party or another, often leaving election outcomes decided by a few votes in pivotal swing states.

The heatmap below shows the democratic vote share in the 50 states and DC for each presidential election since 1980 with darker blue indicating a higher vote share for Democrats and a darker red indicating higher support for Republicans. We can see increasing consistency over time in which party each state supports as well as increasing shades of blue and red among these states, showing increasing polarization as states vote increasingly for a single party.

This dynamic of close elections and polarization creates a host of problems, but one critical problem it creates for national security. With elections decided by a few votes in a few states, it leaves things more vulnerable to hacking by bad actors because small changes in these areas can swing the election.

This hacking can take one of two main forms: (1) deliberately altering vote counts through manipulation of voting machines or ballots and (2) indirectly affecting votes through spreading misinformation that affects voter decisions and turnout.

While we have not seen an example of the first kind of hacking, the 2016 election brought about a serious example of the second. Russia attempted to influence the election and hurt the Clinton campaign while boosting Trump’s chances. They directly hacked DNC servers to steal emails and leak harmful information while also producing and releasing fake news on social media platforms. Both seemed to have a goal of increasing turnout among conservative voters and increasing support for Trump.

Understanding the potential impact of this indirect vote manipulation is difficult because we can’t observe the counterfactual of the 2016 election without interference and compare results. We can though look at how effective manipulation like this can be by building a model to simulate the 2016 election and comparing outcomes in a world with and without manipulation.

Our model will be super simple though it will capture the important aspects. We will take 2016 as the base case and fix election results for states with more than a two percent vote margin, meaning the results of states where either candidate won by more than two percent stay the same.

For states where either candidate won by less than two percent, we will simulate the winner treating them as independent coin flips. While we know states do not vote independently of one another, adding in the correlation does not dramatically change the results and just complicates things unnecessarily, so this assumption is fine.

From 2016, the five states we will simulate results of are Florida, Pennsylvania, Michigan, Wisconsin, New Hampshire, and Minnesota. Ignoring these states, Trump won 232 electoral votes from “safe states” while Clinton won 218 electoral votes. Intuitively, bad actors are incentivized to target states they a priori know will be close since these states matter more in the Electoral College and are easier to change.

Under no manipulation, this simple model puts the odds of Trump winning the election at 63%, which seems roughly reasonable given the actual results.

We can accommodate potential manipulation by simulating the five close states above from a biased coin (probability Republicans win > 50%) instead of from a fair coin (probability Republicans win = 50%). This is because increasing turnout among one side’s voters or shifting independents and/or undecideds will effectively make it more likely for that party to win the state. We run this calculation for various “levels of manipulation” proxied by how biased the coin we use for simulations is (how far above 50% it is).

We run simulations for various levels of manipulation. Each percent of manipulation across these five states results in roughly a 1.6% change in win probability.

A question becomes how much can advertising skew elections?

Research on the effectiveness of Fox News has shown that it can swing vote shares by half a percentage point and also affect turnout. While Fox News is quite influential due to its reach and scale, so is social media which is becoming an increasingly important news source for many people. A campaign on the scale of Fox News would have changed would have been enough to change the results in Michigan and New Hampshire for example.

As bad actors refine their strategies and become much more effective at mobilizing voters and feeding them false information, it isn’t inconceivable to have them changing the win probabilities in small states by five percent or more, skewing election by more than that.

Research has consistently shown that people are predisposed to like favorable news for their side and are motivated by dislike of the other political party. Voter psychology combined with increasingly close elections and the weird Electoral College system leaves voters prone to manipulation. The current climate creates a real incentive for a repeat of what we saw in 2016 and poses a real national security problem.

--

--

Politics, Policy, Economics, & Data

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Vinod Bakthavachalam

I am interested in politics, economics, & policy. I work as a data scientist and am passionate about using technology to solve structural economic problems.